4.2.1 Exclusions
We now simulate two alternative policies, where exclusions are provided for owner-occupied housing and noncorporate equity. These exclusions are made in our model by subtracting (1−κdur)hoj and ωnwt aj respectively from the wealth tax base in equa-tion (3.1), where ωtnw is the endogenous and time-varying portfolio share of financial assets held in the form of noncorporate equity. Holding constant the top-1% threshold of y¯ = $3.249 million, we internally calibrate the tax rate in these two alternative sce-narios so that static revenue-consistency with the benchmark policy is maintained. This
34Consumption of housing services is optimally chosen in proportion to consumption of non-housing market goods in our model; households attempt to smooth both over their lifecycles. See Appendix A.
is achieved at τw = 0.0155 and τw = 0.0119 for the noncorporate exclusion and housing exclusion policies respectively. As in the benchmark policy, all revenue raised from a given policy change is used to pay down federal government debt for the first 40 years following implementation.
Effect on Household Wealth: The deccumulation of household wealth held in the excluded asset class is attenuated under each alternative policy as shown in Figure 1.
Relative to the benchmark policy, aggregate housing is 0.3% larger on average over three decades under the housing exclusion, while aggregate deposits are 0.7% larger under the noncorporate equity exclusion.35 Each exclusion policy generates endogenous avoidance behavior where relatively more wealth is held in the preferential asset class. When housing is excluded, wealth taxpayers36 hold 6.9% more housing on average than is held under the broad-based policy. Similarly, when noncorporate equity is excluded from the tax base, wealth taxpayers save 4.7% more on average than under the benchmark. Other households respond to price effects: more deposits from high-wealth households implies a smaller increase in the portfolio rate of return, and these lower-wealth households increase savings by less than they do under the benchmark. With a smaller increase in permanent income, the households also increase housing by less.
Effect on Productive Activity: While the time paths of the private factors of pro-duction are relatively uniform across sectors when housing is excluded from the wealth tax, they differ significantly when noncorporate equity is excluded, as shown in Figure 2.
Only the exclusion for noncorporate equity distorts the financial intermediary’s portfolio allocation decision: with relatively cheaper equity to finance operations, the noncorpo-rate sector expands while the corpononcorpo-rate sector shrinks, consistent with the findings of Alvaredo and Saez (2010). This reallocation of economic activity amounts to a 1.8 per-centage point increase in the noncorporate sector’s share of total output (from 29.1%
to 30.9%) after three decades. Because this sectoral shift acts as a drag on total tax revenue (discussed below), a relatively higher path of public debt puts upward pressure on the firm borrowing rate and weakens the crowding-in effect from debt reduction. Con-sequentially, the aggregate capital stock remains 0.3% below its baseline level after three decades under this policy alternative despite nearly reverting to baseline under the other policies. Because of differences in aggregate capital-labor substitution across each policy, however, the paths of aggregate output do not significantly differ.
35Relative to the benchmark policy, aggregate deposits differ by less than 0.03% on average over three decades under the housing exclusion, while aggregate housing differs by less than 0.1% under the noncorporate equity exclusion.
36Our ‘wealth taxpayer’ group remains constant across policies for consistency.
Effect on Tax Revenue:37 Figures 3 and 4 show that while the paths of tax revenue under the housing exclusion differ only negligibly from the broad-based policy, signifi-cantly less tax revenue is raised when noncorporate equity is excluded from the wealth tax. Table 7 shows that annual revenue raised directly from the wealth tax is about $10 billion less under the noncorporate equity exclusion policy than the broad-based policy in the first year, a figure which grows to about $29 billion in the thirtieth year. This occurs because the reallocation of economic activity from the corporate sector to the noncorporate sector substantially reduces corporate income tax revenue while only mod-erately increasing noncorporate income tax revenue. The insufficient offset results in $27 billion less total tax revenue collected on average over three decades, and a cumulative (undiscounted) thirty year revenue cost for excluding noncorporate equity of about $801 billion in 2018 dollars.
4.2.2 Evasion
Recent empirical studies emphasize that, in addition to legal avoidance, illegal evasion via the under-reporting of assets and/or over-reporting of liabilities is an important com-ponent of the overall household behavioral response to wealth taxation (Seim (2017), Durán-Cabré et al. (2019), and Brülhart et al. (2019)). PWBM (2019), PWBM (2020), and Diamond and Zodrow (2020) incorporate evasion into their macroeconomic analyses of wealth tax proposals using a simplified reduced-form approach, whereby households misreport taxable wealth according to an exogenous semi-elasticity.38 To draw contrast with the avoidance behavior highlighted in this paper, we simulate our broad-based pol-icy while allowing for evasion using the same reduced-form approach. This involves the respecification of equation (3.1) to:
Ttw(hoj, aj) = (1 +ετw) max τw(aj+ (1−κdur)hoj −y),¯ 0
where ε is the semi-elasticity of reported wealth with respect to the tax rate. Following previous studies, we choose a value of ε=−13for our simulation.39
Figure 1 shows that the decummulation of household wealth is relatively attenuated for both asset classes with the under-reporting wealth.40 Under the assumption that unreported assets remain productive,41 savings fall by less, and firms are able to borrow at a relatively lower rate. Absent sector-specific financial distortions, the capital stock
37All dollar figures are in 2018 dollars.
38Rotberg and Steinberg (2021) allow for endogenous evasion responses that vary across households.
39This is the central estimate chosen by PWBM (2019) in their review of existing estimates.
40While Brülhart et al. (2019) points out that financial assets are under-reported at a greater fre-quency than housing assets, we assume uniform evasion rates to maintain simplicity and consistency with previous analyses.
41This assumption is maintained in PWBM (2019), PWBM (2020), Diamond and Zodrow (2020), and Rotberg and Steinberg (2021).
recovers from its initial decline to reach a level 0.2% above baseline after three decades, both in the aggregate and across sectors, as shown in Figure 2. As firms begin to increase labor input along with capital, output reaches about 0.2% above baseline at the end of three decades.
Figures 3 and 4 show the time paths of wealth tax revenue and tax revenue from other sources. When evasion occurs at our specified intensity, revenue raised directly from the wealth tax is relatively lower than the benchmark policy by $45 and $51 billion in the first and thirtieth years following implementation (in 2018 dollars), differentials larger than any other policy alternative. Total revenue collected varies less however;
table 7 shows that the $159 billion raised annually on average with evasion is remarkably similar to the $157 billion raised on average each year under the noncorporate equity exclusion. In spite of different implications for the allocation of household wealth and pattern of productive activity, under-reporting of wealth at this intensity implies a similar cumulative thirty-year revenue cost to providing an exclusion for noncorporate equity.